RESEARCH DISCIPLINE
Statistical Analysis for Neuropsychology & Neuroscience Research
Consulting for Brain & Behavior Studies
Neuroscience statistical analysis for EEG and imaging recordings, neuropsychological test batteries, patient studies and predictive models, with participants kept as the unit of inference, multiple testing controlled and results prepared to the standard your field expects.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
Statistical Analysis for Neuropsychology & Neuroscience Research
Neuroscience statistical analysis is the statistical design, analysis and reporting of studies that relate brain measures to behavior, disease or cognition, using neuropsychological test scores, EEG or MEG recordings, imaging, and recordings from neurons or cells. A single participant yields thousands of correlated measurements, such as voxels, electrodes, time points or trials, while the number of participants is often small, so a workflow that tests every measurement as if it came from an independent person and reports whatever crosses a threshold produces false positives and unstable estimates. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides neuropsychology data analysis through in-house PhD-level statisticians, who settle the unit of inference, the strategy for many comparisons and the validation of any predictive model with you before the first test is run.
WHEN TO ENGAGE
Statistical Support at Different Stages of Neuropsychology & Neuroscience Research
Researchers usually contact us at a recognizable point, from the planning of a recording protocol to a reviewer's question about a finished map.
Planning a Recording Study You are designing an EEG, imaging or test-battery study and need the sample, the contrasts, the preprocessing plan and the strategy for many comparisons settled before the first recording.
Analyzing Brain Measures Recordings or scans are preprocessed, and the model for trials and participants, the correction across features and the handling of excluded data need to be chosen, run and checked.
Building a Predictive Model You want to classify patients or predict an outcome from imaging, electrophysiology or test scores in a small sample, and the validation has to show how the model would perform on new participants.
Answering Reviewers A reviewer has asked about the correction for multiple comparisons, the independence of trials or cells, or whether the accuracy of a classifier was estimated on the data used to build it.
RESEARCH CONTEXT
Research Designs and Data We Work With in Neuropsychology & Neuroscience
The datasets behind most brain research statistics hold many measurements per participant, and the nesting of those measurements decides which inference is valid.
01
Neuropsychological Test Batteries
Scores in several cognitive domains for patients and comparison groups, with normative data, practice effects at retest and small clinical samples.
02
EEG and MEG Recordings
Electrode or sensor by time-point data from many trials per participant, where each recording yields thousands of correlated measurements.
03
Structural and Functional Imaging
Voxel, region or network measures from scans, with thousands of tests per person and group comparisons that rest on a modest number of participants.
04
Patient and Lesion Studies
Groups defined by a condition, a lesion or an injury, with small samples, mixed causes and comparison with matched controls.
05
Single-Unit and Cell Recordings
Many neurons or cells recorded from each animal or participant, so the units are nested within subjects and are not independent samples.
06
Longitudinal Cognitive Follow-Up
Repeated testing of patients or aging samples, with practice effects, dropout related to decline and change scores to read against retest reliability.
METHODOLOGICAL CONSIDERATIONS
Statistical Issues We Consider in Neuropsychology & Neuroscience Research
Before any region, window or score is reported, your unit of inference, your correction and your validation are examined, because each can change a finding more than the choice of test.
Multiple Comparisons Across Space and Time Thousands of voxels, electrodes or time points are tested together, so the error control is chosen for the claim, whether family-wise, cluster-level or false discovery rate, and its limits are stated.
Participants as the Unit of Inference Trials, voxels and cells within one participant are not independent participants. Inference about people uses participant-level or mixed models, and cells from one animal are not counted as separate subjects.
Overfitting in Small Samples Models with many features and few participants overfit. Feature selection and tuning stay inside cross-validation, and folds are split by participant so that no person appears in both training and test data.
Preprocessing Decisions Choices in artifact rejection, filtering and normalization can change a result. They are fixed before group comparisons, recorded, and checked for their influence on the estimates.
Retest Reliability and Real Change A change in a test score is read against the test's retest reliability and practice effect, so that reliable change can be separated from measurement noise.
Regions Chosen After the Data Regions or time windows selected after seeing where an effect appears give inflated estimates. Analyses are marked as prespecified or exploratory, and every estimate carries its interval.
STATISTICAL METHODS
Methods Selected for the Research Question and Study Design
Each method family follows the data structure and the number of participants in your study, and the line under each title says why it suits brain and test data.
Longitudinal & Multilevel Analysis
Trials within participants and neurons within animals are nested levels, so mixed models give group effects the standard errors that the number of participants supports.
Bayesian & Computational Methods
Permutation tests, run as Monte Carlo resampling, control error across thousands of features without a parametric map model, and bootstrapping resamples participants, not trials.
Comparative & Hypothesis Testing
Patient and control groups are compared as a difference with its interval, with rank-based tests for small or skewed scores and covariates such as age and education in the model.
Regression & Predictive Modeling
Models that predict a diagnosis or a score keep the predictors few relative to the participants, and report discrimination and calibration from validation on held-out people.
Multivariate, Scale & Reliability
Test batteries are checked for reliability and structure, and many correlated measures are summarized into a few components before they are related to a diagnosis.
Clinical, Survival & Epidemiological
A new test or biomarker is compared with a clinical diagnosis, and sensitivity, specificity and ROC analysis show how well it separates patients from controls.
DELIVERABLES
Statistical Outputs Prepared for Your Research
The outputs follow the structure of a brain study, from the review of recording files and preprocessing logs to the scripts that regenerate every figure.
Data & Study Review Recording or test files, trial and subject identifiers, preprocessing logs and the data dictionary are reviewed, so every value can be traced to a participant, a trial and a time point.
Statistical Analysis The agreed mixed, permutation or predictive models, with the multiple-comparison strategy, participant-level cross-validation and the planned sensitivity analyses.
Effect Estimates & Uncertainty Group and condition effects with confidence intervals, cluster-level or corrected inference stated for what it supports, and accuracy with its interval for any predictive model.
Tables & Figures Group score tables, effect maps or time-course plots with corrected thresholds marked, ROC and calibration plots, and the participant counts after each exclusion.
Statistical Interpretation Results are explained within what the inference supports, such as a cluster and not a single voxel, and findings from small samples are described as estimates that need replication.
Analysis Files R scripts, Python code, SPSS syntax or GraphPad Prism project files, with preprocessing choices and exclusions documented so that each figure can be regenerated.
REPORTING
Reporting Neuropsychology & Neuroscience Research
The standard depends on the type of study you report, and each is applied only where it fits.
Studies are reported to APA JARS-Quant. Imaging and electrophysiology studies follow the OHBM COBIDAS recommendations, which ask for the acquisition, the preprocessing and the multiple-comparison correction to be specified in enough detail to repeat them. Cohort and case-control studies report to STROBE, a new test evaluated against a clinical diagnosis to STARD 2015, and a model that predicts a diagnosis or an outcome to TRIPOD+AI. Exclusion counts, the correction used for each map and the distinction between prespecified and exploratory analyses are tabulated from the analysis itself, and the instructions of your target journal are checked last.
ILLUSTRATIVE EXAMPLE
How the Analysis May Look in Practice
An illustration of how a small recording study moves from a plan to a defensible, reportable result.
ILLUSTRATIVE EXAMPLE · MEMORY CLINIC NEUROPHYSIOLOGY RESEARCH · UNITED STATES
A neuropsychology group recorded EEG during a memory task in patients and matched controls, and asked when and where the groups differed and whether a classifier could separate them.
The sample was small and each participant contributed many trials. Before analysis, we fixed the plan in writing: preprocessing and artifact rules recorded in advance, the participant as the unit of inference, and a cluster-based permutation test across electrodes and time points to control the family-wise error rate, interpreted at the level of the cluster and not as the exact onset or location of the effect. For classification, feature selection and tuning were placed inside a nested cross-validation with folds split by participant, so that no person contributed to both training and testing, and a simple model was preferred because of the sample size.
Outcome: The group received the cluster-level permutation result with the cluster's extent across electrodes and time, the cluster average shown descriptively and marked as selected by the test, the cross-validated accuracy with its interval and a calibration plot, a TRIPOD+AI checklist, a methods paragraph and the scripts.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
RESEARCHER VOICE
What Researchers Say
Feedback from a neuroscience researcher on a small sample and unstable estimates, and on a concise output the wider team could use.
“We contacted OnlineSPSS because our dataset was not a textbook example and we needed someone familiar with Neuropsychology & Neuroscience. They focused immediately on a relatively small sample and unstable estimates, then built the analysis around the actual study design. The final output was concise and easy to use with the wider team.”
HOW WE ASSESS YOUR PROJECT
We Review Your Actual Research
Your assessment is based on a review of your actual research materials and analytical requirements, not a generic project category.
You Submit Your Project Details
Share whatever is available at the current stage, such as the research aim, protocol or proposal, dataset, variable list, manuscript or statistical requirements.
We Review Your Research and Define the Scope
Your statistically trained Research Coordinator reviews the materials to understand the study design, research questions, data structure, analytical requirements and expected deliverables. If clarification is needed, additional documents or information can be requested before the assessment is finalized.
Your Analysis Is Completed and Delivered
Once the scope and fixed project fee are agreed, the in-house statistician completes the work and the Research Coordinator remains the communication point through delivery, with the analysis, deliverables and timeline as agreed.
SOFTWARE
Statistical Software for Neuropsychology & Neuroscience Research
Neuroscience analyses are run in the software your laboratory already uses, with syntax or scripts provided where applicable.
Statistical Support Within Your Research OnlineSPSS provides the agreed methodological, analytical, and statistical reporting support. The researcher or research team retains the scientific question, substantive interpretation, conclusions, and overall responsibility for the research.
FREQUENTLY ASKED QUESTIONS
Questions About Statistical Analysis for Neuropsychology & Neuroscience
Short answers to what researchers ask before sharing brain data, from multiple comparisons and nested trials to validation, confidentiality and the assessment.
What statistical methods are used in neuroscience research?
Neuroscience research most often uses mixed models for trials within participants, permutation and cluster-based methods for maps of many features, rank-based or exact tests for small groups, and ROC analysis and validated prediction models for classification. The choice depends on the number of participants, on how many measurements each contributes and on the claim the study wants to make. Our Statistical Methods pages show how each family is applied.
How is multiple testing handled in imaging or EEG data?
The correction is chosen for the claim you want to make, because family-wise control, cluster-level inference and false discovery rate each support a different statement about where or when an effect occurs. Cluster-level inference supports an effect somewhere within a cluster, not exact edges, and false discovery rate controls the expected share of false findings among those flagged. The method and threshold are fixed before the test.
Why can't each trial or cell be treated as a participant?
Trials, voxels and cells from one participant or animal share that individual’s physiology, so they are correlated and are not independent samples. Counting them as separate subjects overstates the sample size and produces false positives. A mixed model with a random effect for participant or animal, or a summary for each participant, keeps the inference at the level of the individuals.
How do you validate a model built on a small clinical sample?
Cross-validation is split by participant, with feature selection and tuning kept inside each training fold, and accuracy is reported with a confidence interval and a calibration check. A small sample gives wide intervals, so a simple model with few predictors is preferred, and validation on an independent sample is the stronger test where one exists.
How is confidential research data handled?
Study datasets, protocols and analysis files are handled confidentially, and project information is not published. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Discuss Your
Neuropsychology & Neuroscience Research With Us
Share your research question, design and available data, even if only a protocol or pilot recordings exist. We review how your measurements are nested and confirm the approach before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014